Look at the cleaning clause in almost any solar O&M contract and you will find the same sentence: the contractor shall clean the modules four times per year. It is easy to write, easy to verify, and it measures nothing anyone actually cares about. Four cleanings performed at the wrong times, in the wrong sequence, with the wrong water, satisfy the contract completely while the plant loses yield all year. Getting cleaning KPIs and SLAs in solar O&M contracts right means paying for the outcome instead of the activity.
Why an activity count is the wrong KPI
A frequency-based clause creates three predictable failures.
It fixes the schedule against a calendar rather than against conditions, so cleaning happens in the month the contract names instead of the month soiling peaks. It rewards speed, because once the count is the deliverable, the contractor’s margin comes from completing each campaign faster. And it gives the owner nothing to enforce when production stays flat afterwards, since the contractor did exactly what was written.
The deeper problem is that it puts the risk on the wrong party. The owner carries the yield risk while the contractor controls every decision that affects it. An outcome-based KPI moves that risk to whoever is actually making the decisions.
KPIs worth writing into a contract
Four measures cover most situations. They can be combined, and the right mix depends on how much of the cleaning decision the contractor controls.
Soiling ratio maintained above a threshold. The soiling ratio, defined in IEC 61724-1, compares the output of a soiled reference device to a clean one. Writing a floor into the contract, for example that the soiling ratio shall not fall below a stated value for more than a defined number of days per year, converts cleaning from a schedule into a performance obligation. It is the cleanest KPI available where measurement infrastructure exists.
Performance ratio with a soiling adjustment. Where the contractor holds broader O&M scope, a PR guarantee already captures cleaning. The trap is a PR clause that excludes soiling from the contractor’s responsibility, which quietly removes the only mechanism that would have made cleaning matter. Check the exclusion list before relying on PR as your cleaning KPI.
Cleaning system availability. Where cleaning is automated, availability of the cleaning equipment is a direct and easily measured KPI: percentage of blocks with an operational system, percentage of scheduled passes completed, time to restore after a fault. This is the natural KPI once cleaning becomes part of an automated program rather than a mobilised service, a shift explored in our article on solar O&M automation.
Event response time. Dust storms, construction nearby, bird activity and harvest season all create step changes in soiling that no schedule anticipates. A response-time obligation, measured from a defined trigger such as a soiling ratio drop of a stated size, covers the cases a periodic KPI misses entirely.
Measuring what you contracted
An outcome KPI is only enforceable if both parties trust the measurement. Agree the method in the contract, not after the first dispute.
- Soiling stations. A paired clean and soiled reference device gives a direct soiling ratio. Specify how many, where they sit, who maintains them and who cleans the reference device, since a soiling station that nobody cleans measures nothing.
- Clean reference rows. Comparing a deliberately maintained clean string against representative production strings is cheaper and works well on larger plants, provided the reference is genuinely representative in orientation and shading.
- Modelled versus actual output. Useful as a cross-check, weak as a contractual basis, because irradiance data, degradation assumptions and availability all enter the calculation and every one of them is arguable.
- System logs. Automated cleaning platforms record every pass by row, with timestamps, water use and faults. That record removes most of what O&M disputes are usually about, which is one of the practical arguments in our overview of remote monitoring in robotic solar cleaning.
Write down the measurement source, the averaging period, the exclusions for snow or rain days, and the reporting format. Most cleaning disputes are measurement disputes wearing a different hat.
Scope boundaries that cause disputes
The clauses that generate arguments are rarely the KPI itself. They are the items nobody assigned.
Water. Who supplies it, who pays for it, what quality is required, and who owns the treatment equipment. A contractor obliged to hit a soiling target with untreated brackish water has a legitimate grievance.
Module damage. Liability for breakage or scratching during cleaning, the deductible, the inspection process and the evidentiary standard. This should be explicit rather than inherited from a general indemnity clause.
Method compliance. A requirement that cleaning stay within the module manufacturer’s documented limits for pressure, water quality, contact materials and temperature, with the contractor responsible for warranty consequences of departing from them.
Site access and curtailment. Cleaning windows, coordination with tracker control, security escort requirements, and who bears the cost when access is denied.
Reporting. Frequency, format, level of detail, and whether raw data is delivered or only a summary. Owners who accept summary reporting lose the ability to audit the KPI they negotiated.
In-house automation versus outsourced crews
The contract shape follows the delivery model, and the two are genuinely different documents.
Outsourced manual cleaning is a service agreement. The commercial logic pushes toward the minimum acceptable frequency, because every additional campaign is mobilisation cost, and the owner pays for activity rather than for condition.
Owner-operated automated cleaning changes the shape entirely. The capital sits with the asset, the marginal cost of an additional pass is small, and cleaning frequency becomes an optimisation rather than a negotiation. The O&M contract then covers maintenance and availability of the cleaning system, not the cleaning itself, and the KPI naturally becomes uptime. The financial comparison between the two models is worked through in our analysis of the ROI of autonomous solar panel cleaning.
A middle model exists, where a provider supplies and operates automated equipment under a performance contract. It combines an outcome KPI with no capital outlay, and it works well provided the availability and end-of-term clauses are drafted carefully.
Penalties, bonuses and getting the incentives right
Penalties are the default and they are only half a mechanism. A penalty tells the contractor what to avoid; it does not tell them what is worth extra effort.
The stronger structure ties both directions to measured energy. If the soiling ratio stays above the threshold, the contractor shares in the additional production. If it falls below, the deduction reflects the estimated lost revenue rather than an arbitrary figure. That alignment is what turns cleaning from a cost line into a yield lever.
Keep the penalty proportionate and capped. A liability that exceeds the whole contract value simply gets priced back into the rate, or filtered out at tender stage by the bidders you most wanted.
Include a review mechanism as well. Soiling conditions change when a neighbouring plot is developed, a road is built or land use nearby shifts. A KPI set on year-one conditions should be revisitable on defined triggers.
A contract checklist
- Outcome KPI defined: soiling ratio floor, PR with soiling included, cleaning system availability, or a combination
- Measurement method, instrumentation, ownership and maintenance responsibility named
- Averaging period, exclusions and reporting format specified
- Event-response obligation with a defined trigger
- Water supply, quality specification and treatment responsibility assigned
- Compliance with module manufacturer cleaning limits as a contractual requirement
- Damage liability, deductible and inspection process stated
- Access windows and tracker coordination agreed
- Bonus and penalty tied to measured energy, proportionate and capped
- Review triggers for changed site conditions
The bottom line
Cleaning is the only O&M activity that touches every module several times a year, and it is usually the one specified in the least detail. A frequency clause buys visits. An outcome KPI buys performance, and it puts the decision-making risk with the party holding the decisions.
Before the next contract renewal, pull the cleaning clause and ask a simple question: if the contractor did everything this document requires and the plant still underperformed, would you have any recourse? If the answer is no, the clause needs rewriting. Solaris Hydrobotics works with owners and O&M providers on the measurement side of exactly that question, since an automated system produces the data an outcome KPI depends on.
Frequently Asked Questions
What is a soiling ratio and how is it measured?
The soiling ratio compares the output of a soiled reference device with a clean one, and is defined in IEC 61724-1. It is measured with a paired soiling station or by maintaining a deliberately clean reference string and comparing it against representative production strings.
Should cleaning be included in a performance ratio guarantee?
It can be, and doing so is often the simplest structure where the contractor holds full O&M scope. The critical detail is the exclusion list: many PR guarantees exclude soiling, which removes cleaning from the guarantee entirely while appearing to cover it.
How many cleanings per year should a contract specify?
Ideally none as a fixed number. A frequency clause locks the schedule to a calendar rather than to conditions. Where a minimum is required for budgeting, pair it with an outcome KPI so that the count is a floor rather than the definition of performance.
Who should pay for cleaning water?
Whoever controls its quality and availability, and the contract should say so explicitly along with a quality specification. Holding a contractor to a soiling target while supplying water that leaves mineral or chloride deposits is not an enforceable arrangement.
How do KPIs change when cleaning is automated?
The KPI shifts from campaigns completed to system availability and scheduled passes completed, because the equipment is on site continuously rather than mobilised periodically. Verification also becomes simpler, since the system logs every pass by row automatically.
What penalty level is appropriate for missed cleaning targets?
One tied to estimated lost revenue rather than an arbitrary figure, proportionate to contract value and capped. Pairing it with an upside share for exceeding the target produces better outcomes than penalties alone, because it gives the contractor a reason to optimise rather than simply to comply.